Toward an ARPEGE-Based Aviation Turbulence Combined Diagnosis

The use of Electronic Flight Bag systems (EFB) on board commercial aircraft enables the overlay of the aircraft flight trajectory and forecasts of turbulence areas, thus raising the pilot expectations of the accuracy of the turbulence areas contours . Some airlines are also considering the promotion of flights with zero turbulence hazard. Thus accurate turbulence forecasts for commercial aviation applications are becoming more important. However, studies suggest that current forecasting procedures will not satisfy these expectations, and that further, that climate change will increase the frequency of turbulence events. A wide set of turbulence predictors (or indices) are already described in the literature. Depending on their definition, the predictors identify different components of the turbulence, but none of them is perfect. The combination of different indices has shown to provide better results, as the GTG (Graphical Turbulence Guidance) developed at NCAR . This method is based on two steps. The first one consists in rescaling every single index to EDR values by using an in-situ airborne EDR climatology distribution [Sharman et al 2014]. The second step is to find an ensemble mean of selected indices which provides the best accuracy based on a 1-year database of turbulence observations (i.e., pilot reports of turbulence or PIREPs) [Sharman et al 2017]. However, the obtained combination depends on the NWP model used and its spatial and vertical resolution.

This study evaluates the adaptation of the GTG methodology to Météo-France’s global NWP model named ARPEGE (horizontal resolution of 0.25° after interpolation). The whole calibration is operated with the NCAR PIREP/in-situ EDR database over the CONUS domain, focused on Clear Air Turbulence indices at upper levels (> FL180). The selected indices are compared depending on the NWP model used: GFS or ARPEGE. The new GTG ARPEGE combination is validated over the CONUS and Europe against commonly used turbulence indices. For this purpose, a database of European aircraft turbulence observations is used over the year 2017. Scores show the improvement obtained by the combination of indices. Other strategies for combining indices are explored, such as logistic regression and Random Forests. Scores, benefits and drawbacks of each method from an operational point of view are presented. The study is completed by a test case analysis over the CONUS.